Conversation Handling
This page covers everything that happens inside an open conversation: reading the thread, composing replies, using AI assistance, managing statuses and assignments, and performing bulk operations across multiple conversations.
The conversation thread
When you open a conversation, the thread shows the full history of all interactions in chronological order. Several different message types appear in the thread:
- Customer messages — messages sent in by the customer through any connected channel
- AI agent messages — replies sent by an AI agent on behalf of your workspace
- Human agent messages — replies composed and sent by a member of your team
- System events — non-message entries such as assignment changes, status updates, and merge actions
Phone calls
When a phone call comes in, a call panel pops up on screen as long as you are logged in. You can choose Answer in Inbox to open the conversation in the inbox, or simply Answer to take the call without navigating away from your current view. Call recordings and voicemail transcriptions are available at the top of the conversation details panel after the call ends.
Escalation handoff card
When an AI agent escalates a conversation to a human agent, a handoff card appears in the thread at the point of escalation. The card gives you everything you need to pick up where the AI left off:
- The reason the AI escalated
- Actions the AI already took before handing off
- How many messages the AI sent during the conversation
- How long the conversation has been active
- The customer’s detected sentiment
- Suggested resolution steps with checkboxes so you can mark each step complete as you work through them
Compose area
The compose area sits at the bottom of the thread. It is a rich text editor where you type your reply. Formatting options such as bold, italics, and links are available in the toolbar above the text field.
Email signatures
When replying to email conversations, the system appends your workspace’s email signature automatically. The signature is shared across all agents in the workspace, but you can use the {user.name} template variable so each reply shows the name of the agent who sent it.
Failed message delivery
If an email or SMS fails to send, a failure indicator appears directly on the message in the thread so you know to retry or follow up through another channel.
Send shortcuts:
- Ctrl+Shift+Enter — send your reply and immediately mark the conversation as resolved
- Ctrl+Shift+S — send your reply and snooze the conversation
Translate
If a conversation is in a language other than your preferred language, you can translate the messages directly in the thread. Translation is batched and processes up to 25 messages at a time. Once translated, each message displays the translated text alongside the original.
Activity tracking
The Help Desk automatically tracks how long you actively spend working on each conversation. The timer starts when you open a conversation and pauses after 60 seconds of inactivity. Sessions shorter than 3 seconds are not counted. This data feeds into agent productivity reports and is visible on the conversation record.
AI assist and macros
Autocomplete
As you type a reply, the system streams AI-generated suggestions in real time beneath your cursor. Autocomplete draws on:
- The messages in the current conversation
- Previous replies that human agents have sent to the same contact
- Replies from other conversations that share the same topic
If you have already started typing, the AI continues from your existing text rather than replacing it. For email conversations, autocomplete automatically generates an appropriate greeting to open the reply.
AI Draft
The Copilot panel, accessible from the sidebar, shows a pre-generated draft response based on the conversation context. You can:
- Edit the draft inline before sending
- Regenerate the draft if the suggestion does not fit
- Copy the draft to your clipboard
- Send the draft directly without copying it first
Macros
Macros are saved response templates that agents can reuse for common replies. To insert a macro, type / in the reply editor to open the macro menu, then use the arrow keys or search to find the one you need.
When you start typing and autocomplete detects a matching macro, it can also return the macro text verbatim rather than generating a new AI response. Macros receive a scoring boost so they surface first when they are relevant to the conversation.
Your workspace administrator manages macros from the workspace settings.
Statuses and resolution
Resolution types
Resolution type describes how a conversation ended:
- Unresolved — no outcome has been recorded yet; the conversation is still open
- Escalated — the conversation was handed to a human agent and is awaiting action
- Soft resolved by AI — the AI closed the conversation after the customer stopped responding
- Hard resolved by AI — the customer explicitly confirmed that their issue is resolved
- Resolved by human — a human agent marked the conversation complete
Conversation statuses
The status displayed in the inbox reflects the current state of the conversation:
- New — escalated but not yet assigned
- Open — escalated and assigned, pending agent action
- Pending — agent has responded, waiting on the customer
The underlying conversation status also determines the color bar shown in the inbox list: amber for New, rose for Open, and blue for Pending. When you send a reply, the status automatically changes from Open to Pending.
Resolving a conversation
You can resolve a conversation in several ways:
- Click Resolve in the conversation thread toolbar to resolve the conversation only.
- Click Resolve and Complete to resolve the conversation and also mark all linked tickets as completed in one step.
- Use the Ctrl+Shift+Enter send shortcut to send your reply and resolve in one step.
Use Resolve and Complete when you are done with both the conversation and any associated tickets. Use Resolve alone if the conversation is finished but the tickets still need follow-up.
When a human agent resolves a conversation, the system can optionally send a CSAT satisfaction survey to the customer. Whether the survey is sent depends on your workspace configuration and the channel type.
Reopening a conversation
To reopen a resolved conversation, use the Unresolve action in the thread toolbar. If your workspace is configured for reassign-on-reopen, the system automatically re-routes the conversation to an available agent.
Snooze
Snoozing temporarily removes a conversation from your active views (if configured to filter out snoozed conversations) without resolving it. Use snooze when you are waiting on something and want the conversation to reappear at a specific time.
To snooze a conversation:
- Click Snooze in the thread toolbar.
- Set the date and time for the conversation to reappear.
- Optionally add a reason to remind yourself why you snoozed it.
Cancel conditions let you configure event-based wake-up rules — for example, unsnooze the conversation automatically if the customer sends a new message before the scheduled time.
Snoozed conversations appear in the Snoozed view in the sidebar. You can unsnooze a conversation manually at any time by opening it and clicking Unsnooze, or let the snooze expire on its own at the scheduled time.
Assignment and escalation
Manual assignment
You can assign a conversation to any agent or group at any time from the details panel on the right side of the screen. Select an agent from the Assignee dropdown or select a group from the Group dropdown.
Auto-assignment
When a conversation is escalated to a human agent, the routing system assigns it automatically. The default routing strategy is round-robin across available agents, with capacity limits to avoid overloading any single agent.
Group assignment
Assigning a conversation to a group routes it to the group rather than a specific person. The system then routes the conversation to an available member within that group according to the group’s routing rules.
Reassignment
You can change the assignee at any time. When you reassign a conversation, the previous agent is removed and the new agent takes ownership.
Escalation flow
Escalation is not limited to the first message. As the AI agent responds to new messages in a conversation, it continuously analyzes whether it can still handle the issue or should escalate to a human. Escalation can happen at any point in the conversation.
When an AI agent escalates a conversation, the system performs the following steps in sequence:
- Marks the conversation status as Escalated
- Applies a topic label to classify the issue
- Optionally converts the channel from webchat to email if your workspace is configured for this
- Routes the conversation to a human agent using the auto-assignment rules
- Creates a linked ticket so the escalation is tracked as a work item
- Triggers any automation flows that are configured to run on escalation
Merging and splitting
Merge
If a customer contacts you about the same issue through multiple channels, you can merge those conversations under a single parent. The merged child conversations remain accessible and are listed under the parent, but all further activity is tracked in one place.
To merge conversations for the same customer:
- From the details panel, open the Timeline
- Select the conversations you want to merge.
- Click Merge.
Unmerge
To handle a child conversation separately again, detach it from its parent using the Unmerge action. The child conversation becomes independent and is no longer listed under the parent.
Priority
Priority is displayed on a 1-to-5 scale:
- No Priority — no urgency assigned
- Low — low urgency
- Medium — standard priority for most conversations
- High — elevated urgency, should be addressed soon
- Urgent — requires immediate attention
You can set priority manually from the details panel. Priority can also be set automatically by routing priority rules based on the channel, contact group, topic, and other attributes. Higher-priority conversations surface first in sorted views.
Topics, intents, tags, and flags
Topics and intents
Topics and Intents are structured classifications used for reporting and routing. A conversation might be labeled Billing with a sublabel of Refund, for example. Labels are set from the details panel or applied automatically by the AI agent.
Tags
Tags are free-form text labels for flexible categorization. You can add as many tags as you need to a conversation. Tags are not predefined — you type them in as you go.
Flags
Flags are predefined markers set at the workspace level. Unlike tags, the available flag options are controlled by your workspace administrator. Flags are applied automatically based on what the AI detects in the conversation, such as negative sentiment or a safety concern.
Flags let your team quickly categorize conversations using a consistent, shared vocabulary and can be used for filtering in views and routing rules.
Spam
Conversations labeled with the Spam topic appear in the Spam view in the sidebar. To label a conversation as spam, set the topic to Spam from the details panel.
CSAT surveys
After a human agent resolves a conversation, the system can automatically send a satisfaction survey to the customer. The survey asks the customer to rate their experience from 1 to 5 stars and optionally leave a comment.
For email conversations, the survey is sent after a configurable delay (the default is 24 hours). Guards in the system prevent duplicate surveys from being sent if the customer contacts you again before the delay has passed.
Surveys are not sent for:
- Webchat conversations (satisfaction is handled inline in the chat widget)
- Test conversations
- Conversations labeled as Spam
- Child conversations that have been merged into a parent
You can toggle whether or not to send a CSAT survey upon resolving a conversation by using the toggle in the Resolve dropdown.
Bulk actions
You can act on multiple conversations or tickets at once without opening each one individually. Select items using the checkboxes in the list view, or hold Shift and click to select a range of items.
With conversations selected, the bulk action bar appears at the bottom of the screen:
- Resolve — mark all selected conversations as resolved
- Set priority — apply a priority level to all selected conversations
- Add flags — add one or more flags to all selected conversations
- Assign — assign all selected conversations to a specific agent
With tickets selected, the following bulk actions are available:
- Mark as completed — move all selected tickets to the Completed status
- Set priority — apply a priority level to all selected tickets
- Assign — assign all selected tickets to a specific agent
Export
To export your current conversation list, apply the filters you want and then click Export. The system downloads a CSV file containing the conversations that match your active filters.
Test vs. live conversations
Test conversations are marked so they do not affect live reporting. Use Test Coverage to create and review these scenarios, then return to the Inbox to see real customer activity.
Next steps
- Details Panel — the sidebar tabs and what they show
- Inbox & Views — filtering, saved views, and search
- Tickets — ticket management and workflows
- Keyboard Shortcuts — all available shortcuts
- Routing & Assignment — how auto-assignment and routing rules work
- Notifications & Mentions — notification types and @mentions